- SCADA-to-MES integration requires three things: a reliable connectivity protocol (OPC-UA preferred), a data-mapping exercise that links SCADA tags to production context, and a validation step before going live.
- The most common failure point isn't the technology — it's mismatched data models. SCADA thinks in machine tags; MES thinks in work orders. Mapping these is a business logic exercise, not just an IT one.
- Read-only integration at the SCADA layer protects control-system integrity — MES should never write back to SCADA without explicit engineering review.
The single biggest mistake in SCADA-MES integration is treating it as an IT project
SCADA-to-MES integration is a business logic problem that happens to have a technical implementation. The technology — OPC-UA, REST APIs, message queues — is well understood. What breaks integrations is the translation problem: SCADA measures physical reality in machine tags; MES measures production reality in orders, batches, and work centres. Mapping one to the other requires production engineers in the room, not just IT.
Every SCADA-MES project that fails does so because someone mapped tags to MES fields without understanding what production context those tags belong to.
Step 1: Connectivity protocol selection
OPC-UA is the right choice for most new integrations. It’s the modern industrial connectivity standard, supports both real-time and historical data access, has built-in security, and is supported by virtually all SCADA vendors.
OPC-DA (the older version) still works for legacy systems but lacks security and is being phased out — use it only if OPC-UA isn’t supported by your SCADA version.
Direct historian access (reading SCADA’s SQL or proprietary historian database directly) works but creates a tight coupling to the SCADA data model that makes future SCADA upgrades painful. Avoid unless OPC is unavailable.
Never configure SCADA-MES integration to write back to SCADA control tags without a formal change management review. MES should read production actuals from SCADA; control setpoints should only flow from SCADA downward to PLCs, not from MES to SCADA. Crossing this boundary can create safety and quality incidents.
Step 2: Tag audit and data mapping
Before writing a line of integration code, produce a tag register: every SCADA tag you intend to bring into MES, its data type, its scan rate, its units, and — critically — its production context: which machine, which process step, which product family does this tag relate to?
Map each tag to the MES entity it belongs to: work centre, resource, operation step. This mapping is what makes raw machine data meaningful at the production level.
Common mapping decisions that require engineering input:
- Which tags define “machine running” vs “machine idle” vs “machine faulted”?
- How do you handle multi-product machines where the same tag means different things for different products?
- Which SCADA events correspond to MES production order start/stop signals?
Step 3: Data validation before go-live
Run the integration in parallel for 2-4 weeks before switching MES to rely on SCADA-sourced data. Compare SCADA-derived production counts against manual counts or existing ERP actuals. If they match within acceptable tolerance, the mapping is correct. If not, the data-mapping exercise needs to be revisited — not the integration layer.
Build a data-quality dashboard as part of the integration — showing tag read success rates, latency, and value ranges. This becomes your early-warning system for connectivity issues before they affect production reporting.
How LeanQubit simplifies this
FactoMES and FactoLake include pre-built connectors for the most common SCADA systems and a visual tag-mapping interface that lets production engineers (not just IT) configure the SCADA-to-MES context mapping. The validation dashboard is built in.
Frequently Asked Questions
Simple environments (one SCADA system, under 500 tags, clear production context): 4-8 weeks including validation. Complex environments (multiple SCADA systems, proprietary historians, multi-product machines): 3-6 months. The data-mapping exercise is the longest part, not the technical integration.
A well-designed integration includes fallback to manual entry when SCADA connectivity is lost. MES should never be fully dependent on real-time SCADA for basic functionality — production operators need to be able to record work even during connectivity interruptions.
Yes — FactoLake supports direct historian database reads for common systems, REST/MQTT for modern IIoT sensors, and custom connector development for proprietary systems. OPC-UA is preferred for new integrations but not a hard requirement.
Related reading: SCADA vs MES vs ERP · How SCADA Fits into an Industry 4.0 Architecture · Using OPC-UA for AI Integration